The problem can be formulated as a non-convex loss minimization problem in statistical learning, and our online algorithms have vectorized and adaptive step sizes to ensure high estimation accuracy in three dimensions which have different magnitudes. In order to keep the algorithm from sticking in ...
Graph learning has been an active research area that finds applications across a number of fields including finance, health care, and social sciences. In this chapter, we present an overview of recent advancements in the area of learning graphs from data, in particular undirected, weighted graphs...
This is the first algorithm-dependent result with reasonable dependence on aggregated step sizes for non-convex learning, and has important implications to statistical learning aspects of stochastic gradient methods in complicated models such as deep learning. 展开 ...
1> Deep Learning 里面每一个单元输出的时候都会用到 Activation function 这个函数常是非线性的。从而...
Non-convex Optimization for Machine Learning (2017) 具有隐凸性或解析解的问题 These slides summarize lots of them. Blind Deconvolution using Convex Programming (2012) Intersecting Faces: Non-negative Matrix Factorization With New Guarantees (2015) The why and how of nonnegative matrix factorization (20...
Foundations and Trends® in Machine Learning(共65册), 这套丛书还有 《Bayesian Reinforcement Learning》《Graph Neural Networks for Natural Language Processing》《Machine Learning for Automated Theorem Proving》《From Bandits to Monte-Carlo Tree Search》《A Survey of Statistical Network Models》 等。
machine learning discrete mathematics convex analysis Editors and Affiliations Department of Industrial and Systems Engineering, University of Florida, Gainesville, USA Panos M. Pardalos Mathematics, National Technical University of Athens, Athens, Greece Themistocles M. Rassias About the editors Pan...
Non-convexoptimizationisubiquitousinmodernmachinelearning:recentbreak- throughsindeeplearningrequireoptimizingnon-convextrainingobjectivefunctions; problemsthatadmitaccurateconvexrelaxationcanoftenbesolvedmoreefficiently withnon-convexformulations.However,thetheoreticalunderstandingofnon-convex ...
A vast majority of machine learning algorithms train their models and perform inference by solving optimization problems. In order to capture the learning and prediction problems accurately, structural constraints such as sparsity or low rank are frequently imposed or else the objective itself is designe...
已在人工智能、应用数学领域TOP期刊如IEEE Transactions on Pattern Analysis and Machine Intelligence(5), IEEE Transactions on Information Theory,IEEE Transactions on Image Processing(2),IEEE Transactions on Geoscience and Remote Sensi...